Identification of speculative bubbles using state-space models with Markov-switching

In this paper we use a state-space model with Markov-switching to detect speculative bubbles in stock-price data. Our two innovations are (1) to adapt this technology to the state-space representation of a well-known present-value stock-price model, and (2) to estimate the model via Kalman-filtering using a plethora of artificial as well as real-world data sets that are known to contain bubble periods. Analyzing the smoothed regime probabilities, we find that our technology is well suited to detecting stock-price bubbles in both types of data sets.